EVOLUTION & DEVELOPMENT 5:1, 9–18 (2003)

Development and of adaptive polyphenisms

H. Frederik Nijhout Department of Biology, Duke University, Durham, NC 27708, USA Correspondence (e-mail: [email protected])

SUMMARY is the primitive character due to discontinuities in the environment that induce only por- state for most if not all traits. Insofar as developmental and tions of what is in reality a continuous . In insect physiological processes obey the laws of chemistry and polyphenisms, the environmental variable that induces the al- physics, they will be sensitive to such environmental vari- ternative is a token stimulus that serves as a pre- ables as temperature, nutrient supply, ionic environment, and dictor of, but is not itself, the environment to which the the availability of various macro- and micronutrients. De- polyphenism is an adaptation. In all cases studied so far, pending on the effect this phenotypic plasticity has on fitness, the environmental stimulus alters the endocrine mechanism evolution may proceed to select either for mechanisms that of metamorphosis by altering either the pattern of buffer or canalize the phenotype against relevant environ- secretion or the pattern of hormone sensitivity in different tis- mental variation or for a modified plastic response in which sues. Such changes in the patterns of endocrine interactions some ranges of the phenotypic variation are adaptive to partic- result in the execution of alternative developmental path- ular environments. Phenotypic plasticity can be continuous, in ways. The spatial and temporal compartmentalization of which case it is called a reaction norm, or discontinuous, endocrine interactions has produced a developmental mech- in which case it is called a polyphenism. Although the mor- anism that enables substantial localized changes in morphol- phological discontinuity of some polyphenisms is produced ogy that remain well integrated into the structure and function by discrete developmental switches, most polyphenisms are of the organism.

INTRODUCTION temperature, nutrition, photoperiod, and so on. These envi- ronmental variables affect the underlying chemical and met- In most organisms a can produce many different abolic processes of development directly, without the inter- . The exact phenotype that is expressed depends vention of a specially evolved mechanism. on the environment in which the organism develops. Pheno- typic plasticity can be gradual or discrete. Phenotypes that change with small gradual changes in an environmental vari- ORIGIN AND EVOLUTION able are called reaction norms. In addition to these continu- OF PHENOTYPIC PLASTICITY ously variable phenotypes, some organisms can develop two or more discrete alternative phenotypes, without intermedi- All phenotypes are believed to be primitively plastic. This is ate forms. This phenomenon is called polyphenism and can because developmental, physiological, and metabolic pro- come about in two ways: either when different members of a cesses are normally sensitive to environmental variables species experience discretely different environments (as in such as temperature, pH, ionic strength, and nutrients. Such the case of a bivoltine insect with discrete generations in dif- phenotypic sensitivity to environmental variables leads to ferent seasons) or when a continuously variable environment what is called a reaction norm: the range of phenotypes pro- induces a discrete threshold-like switch from one develop- duced by a given genotype when exposed to a range of values mental pathway to another. of a single environmental variable. Given that phenotypic The mechanisms that mediate these two types of pheno- plasticity is obtained gratis, as a by-product of the physics and typic plasticity are beginning to be understood. As we will chemistry of development, evolution of this plasticity can see below, the development of alternative phenotypes in re- occur in two directions: One results in stabilization of the action norms and polyphenisms can be caused by specially phenotype, effectively eliminating the plasticity, whereas evolved mechanisms that are regulated by variation in the the other results in the exploitation of the plasticity. If this patterns of hormone secretion. Reaction norms can also re- phenotypic sensitivity to environmental variation reduces sult from the fact that the rates and timing of developmental fitness, then it will be eliminated by the evolution of mecha- processes are affected by such environmental variables as nisms that somehow reduce the sensitivity of the phenotype

© BLACKWELL PUBLISHING, INC. 9

10 EVOLUTION & DEVELOPMENT Vol. 5, No. 1, January–February 2003

Table 1. Examples of differences between the selective internal environment by means of feedback mechanisms that environment to which a polyphenism is an adaptation, return the system to a set point. Homeostatic mechanisms in and the inductive environment that actually triggers physiology and metabolism have been studied for a long the polyphenic developmental switch time and are now well understood. In development there has Selective agent (to also been an evolution of mechanisms that buffer the emer- which the polyphenism gent phenotype against environmental and genetic variation. Polyphenism is an adaptation) Inducing stimulus Depending on one’s perspective, the evolved insensitivity to Seasonal Lethal temperature, Photoperiod, variation is called canalization (the ability to return to a de- food scarcity nonlethal velopmental trajectory after a disturbance) or (the temperature Phase () Food quantity/quality Crowding, simple ability to tolerate or be insensitive to environmental temperature, or genetic variation) (Nijhout and Davidowitz 2002). Unlike photoperiod physiological homeostasis, the mechanisms that confer ro- Phase (locusts) Food quantity/quality Crowding Phase (caterpillars) Food quality bustness in development are not well understood at present. Wing length Food quantity/quality Crowding, Evolution has not invariably proceeded to stabilize the photoperiod phenotype in the face of environmental variation. Instead, in Horn length Mating success Food quantity/ many cases phenotypic plasticity has been exploited as a quality Caste (, soldiers) Food quantity/quality, Food quality, mechanism that enables an organism to develop different predators phenotypes that may be adapted to two or more environ- Caste (ants, gynes) Reproduction Pheromones, ments, without requiring the evolution of a genetic polymor- overwintering Caste (bees) Reproduction Nutrition, phism. Thus, rather than having a single phenotype that is pheromones well adapted to a single niche or function, organisms can de- Diapause Lethal temperature Photoperiod, velop alternative phenotypes, each of which may be opti- nonlethal temperature mized for fitness in a different environment. Phenotypic plas- ticity can therefore be divided into two kinds: type 1, which comes gratis in the absence of a homeostatic mechanism that buffers the phenotype against environmental variation dur- to a particular environmental variable. A broad diversity of ing development but that is unlikely to be adaptive, and type homeostatic mechanisms have evolved that buffer physiologi- 2, which is an adaptation to a particular set of environments. cal and metabolic systems in the face of environmental vari- Both type 1 and type 2 plasticity can be manifested as a ation. Homeostatic mechanisms serve to maintain a constant reaction norm, but their underlying developmental physiol-

Fig. 1. Polyphenisms can come about in two different ways: (1) from a reaction norm when the environment is either discontinuous or is only sampled at discrete times or places so that the environment is effectively discontinuous; or (2) from switches in developmental pathways that produce a discontinuous reaction norm.

Nijhout Development and evolution of adaptive polyphenisms 11 ogy and evolutionary history are likely to be very different. Our understanding of the developmental mechanisms that It is unlikely that type 1 plasticity can produce an array of underlie phenotypic plasticity comes almost entirely from phenotypes, each of which just happens, by chance, to have studies on the development of polyphenisms. Polyphenisms the highest fitness in the particular range of environments are discrete alternative phenotypes that develop in response to that induce them. The evolution of adaptive phenotypic plas- environmental variation. Among the best known polyphen- ticity must involve changes in development that alter the isms are the castes of social insects, the alternative seasonal type 1 reaction norm. Reaction norm evolution and the evo- forms of insects, heterophylly in plants, predator-induced lution of adaptive phenotypic plasticity have been the subject polyphenisms in cladocerans, male mating polyphenisms in of numerous theoretical and empirical studies (Via and Lande beetles, phase polyphenisms of migratory locusts, and the 1985; Lively 1986; West-Eberhard 1989; Gomulkiewicz and long- and short-winged dispersal polyphenisms of many in- Kirkpatrick 1992; Via et al. 1995; Schlichting and Pigliucci sects (Cook 1968; Wheeler 1986; Harvell 1990; Nijhout 1998), and a great deal is known about the conditions under 1994; Dingle and Winchell 1997; Schlichting and Pigliucci which the shape of the reaction norm can evolve under selec- 1998). tion in different environments. By contrast, relatively little is Polyphenisms are adaptations to reliable and predictable known about the developmental mechanisms that produce a variations in the environment. From an evolutionary perspec- particular reaction norm nor about the way those develop- tive, perhaps the most interesting thing about a polyphenism is mental mechanisms change when reaction norms evolve. that the inducing environment is not the same as the selective

Fig. 2. (A–D) Temporal pattern of variation of ecdysteroid titers in the hemolymph and of ecdysone (EcR) isoform expression in different tissues of larvae of the tobacco hornworm, (Nijhout 1999).

12 EVOLUTION & DEVELOPMENT Vol. 5, No. 1, January–February 2003 environment. This is to say, the environment to which the al- the fact that a finite period of time is needed to develop a par- ternative phenotype is an adaptation is not the same as the ticular phenotype. In modern polyphenisms the time delay is environment that induces the development of that phenotype likely to be part of the adaptation that ensures that a predic- (Table 1). For instance, a seasonal polyphenisms may be an tive environmental stimulus is sampled for a long enough pe- adaptation to cold, or nutrient stress, but is typically induced riod and at a time when its predictive power is strong. by a change in photoperiod. A change in photoperiod (the Many insects have evolved a well-defined critical period relative lengths of day and night) does not in itself constitute in development when the individual is sensitive to inducing an unfavorable environment, but it is an excellent predictor stimuli, and this critical period occurs long before the alter- of a seasonal change and is thus a good predictor of future native phenotype actually develops. Evolution of the use of temperature or nutrient stress. Because the inducing environ- a token stimulus requires a correlation between the token and ment is a token, or predictor, there is usually a considerable the selective environment and that the organism is somehow time lag between the inducing environment and the selective able to use the token to accurately predict the future environ- environment. The time interval between the sensitive period ment and match its phenotype accordingly. The general condi- for induction and the actual development of the alternative tions required for the evolutionary maintenance of a polyphen- phenotype was initially probably a simple consequence of ism, once it is established, have been studied by Moran

Fig. 3. Endocrine mechanisms underlying the polyphenic switch in insects. The up- per diagram depicts the developmental period. At some time during larval life there is a sensitive period during which specific environmental stimuli such as temperature, photoperiod, or pheromones (Table 1) are integrated. These result in a reprogramming of an endocrine mechanism just before or during metamorphosis and leads to the initiation of one of two alterna- tive developmental pathways that result in different adult phenotypes at metamorpho- sis. Four kinds of hormonally controlled de- velopmental switching mechanisms have been identified to date, illustrated in the bottom four panels, with examples of the polyphenisms in which they have been documented. In all cases, act during tissue-specific sensitive periods. Alternative developmental pathways en- sue depending on whether the hormone is above or below a threshold value during such a period (Nijhout 1999).

Nijhout Development and evolution of adaptive polyphenisms 13

(1992). Little or no research has been done on the evolution- the environment or by manipulating the underlying develop- ary origins of polyphenisms. mental physiological mechanisms that produce the threshold It is likely that polyphenisms originate from continuously (Windig 1994). It is therefore worth examining the mechanism plastic phenotypes. Most modern polyphenisms are, in fact, by which development can be made sensitive to an environ- reaction norms, although this is not always easy to detect in mental gradient so that a threshold phenotypic response results. nature. In many and perhaps most polyphenisms, the discrete alternative phenotypes develop either because the environ- ment is discontinuous or because the environment-sensing DEVELOPMENTAL SWITCHES physiology has a threshold (Fig. 1). A discontinuous envi- ronment is one that would be experienced by a bivoltine insect In all cases where the developmental mechanism of polyphenic (an insect that has two generations per year). In such animals, regulation has been elucidated, the developmental switch that each generation develops in a different season and thus expe- leads to alternative phenotypes is regulated by hormones. riences a different combination of photoperiod, temperature, Virtually all of our knowledge of these mechanisms comes nutrition, and population density. In many cases when such from the studies on the regulation of postembryonic develop- polyphenic insects are exposed to intermediate environmental ment in insects. Because the external characteristics of in- conditions, they develop a range of intermediate phenotypes sects are part of their (nonliving) cuticle, the expression of not normally seen in nature (Nijhout 1994). polyphenism typically requires a molt, and the alternative Examples of threshold environmental sensitivity can be phenotype is expressed in the new cuticle that is synthesized found in the seasonal polyphenisms of multivoltine insects at that time. Thus, depending on the prior environment, a that have a critical photoperiod for the induction of the polyphenic insect can molt into one of several alternative polyphenism (Nijhout 1994). Such insects develop alterna- forms. Insects express both larval and adult polyphenisms. tive phenotypes depending on whether the photoperiod they The latter are associated with the metamorphic molt, and experience is longer or shorter than the critical day length. such insects can facultatively metamorphose into alternative Even with such thresholds it is possible to obtain a range of adult forms. In adult polyphenisms, the environment-sensitive intermediate phenotypes (see Fig. 4), either by manipulating period occurs sometime during larval life. The exact timing

Fig. 4. Phenotypic plasticity in the seasonally polyphenic Araschnia levana. In nature, different generations of this species de- velop in discretely different environments, and this results in a purely diphenic polyphenism represented by the two extreme forms out- lined by boxes. The intermediate forms can be produced by timed ecdysone injections or by intermediate environments and illustrate that a reaction norm lies at the base of this polyphenism.

14 EVOLUTION & DEVELOPMENT Vol. 5, No. 1, January–February 2003 and duration of environment sensitivity can be early, middle ism. Evolutionary adaptation of the alternative morphs of a or late in larval life, depending on the species. polyphenism is most likely facilitated by the fact that hormone- The developmental switch mechanism operates sometime sensitive periods are time and tissue specific, so that devel- during the instar that precedes the molt to the polyphenic opmental regulation is effectively compartmentalized in form and has the following general structure. During the rel- both time and space. Hence, different parts of an organism evant larval instar there are one or more relatively brief peri- could vary and respond to selection independently. The ge- ods during which hormones can alter the course of subsequent netic correlation structure of polyphenic phenotypes, how- development. These are called critical periods or hormone- ever, has yet to be studied in detail. sensitive periods, and they act essentially as binary switches, with alternative developmental pathways being selected de- pending on whether a hormone is above or below a threshold REACTION NORMS value. Interestingly, the principal hormones that control polyphenic development are the same ones that control molt- Perhaps the most interesting thing about having a hormonal ing and metamorphosis, namely, ecdysone and juvenile hor- regulation of development is that development comes under mone. In addition several neuroendocrine hormones have the control of the central nervous system. This is because the evolved specialized associations with polyphenic regulation developmental hormones are directly regulated by neuro- in several species of insects (Nijhout 1994). secretory factors or are themselves neurosecretory hormones. The way in which developmental hormones regulate al- The central nervous system can integrate information about ternative patterns of gene expression is best understood for the hormone ecdysone. The ecdysone receptor is a het- erodimer (Talbot et al. 1993) that acts as a , and different combinations of the alternative subunits of the dimer regulate different patterns of gene expression (Truman et al. 1994). Different isoforms of the ecdysone receptor are expressed in tissue-specific temporal patterns (Fig. 2), and the tissue-specific periods of sensitivity to a particular effect of the hormone is determined by when and where these re- ceptors are expressed. When a peak of hormone secretion coincides with recep- tor expression in only one tissue, then only that tissue will re- spond to the hormone, and the remaining tissues will continue in their current state, as if no hormone had been present at all. Receptor expression, and therefore the hormone-sensitive pe- riods, are regulated independently from fluctuations in hor- mone secretion. All known cases of polyphenic switching of developmental pathways occur by environmentally induced changes in either the timing of hormone secretion, the timing of a hormone sensitive period, or the threshold of hormone sensitivity (Nijhout 1999). The array of possibilities, to- gether with examples of each type of regulation, is illustrated in Figure 3. Different patterns of gene expression result, de- pending on whether or not a peak of hormone secretion co- incides with a period of hormone sensitivity, and this leads to the development of alternative phenotypes. The origin of polyphenisms can be understood in devel- opmental terms as being due to the origin or loss of a coin- cidence between a peak of hormone secretion and a hor- mone-sensitive period. Variation in the timing of hormone Fig. 5. Horn size allometries in two rhinoceros beetles. The al- secretion or receptor expression (or of the threshold of hor- lometry of Phanaeus is a continuous sigmoid curve and suggests mone sensitivity) could produce an occasional partial or full an underlying continuous size-dependent generative mechanism. The allometry of Chalcosoma appears discontinuous. Discontinu- mismatch, which then results in new phenotypes. Genetic ous allometries come about through reprogramming of develop- stabilization of the mismatch in response to some environ- ment and are not reducible to a smooth reaction norm (Emlen and mental signals (but not others) could then fix the polyphen- Nijhout 2000).

Nijhout Development and evolution of adaptive polyphenisms 15 the animal’s internal and external environment and use this in- the alternative forms of a polyphenism differ only in quanti- formation to regulate the secretion of hormones. In this way, tative and not in qualitative ways. development can become responsive to a wide diversity of en- In a few known cases, there is a discontinuity in the tran- vironmental signals, without the need to have developmental sition between one polyphenic morph and another (Fig. 5). processes themselves be sensitive to the environment. In such cases the alternative morphs appear to have different All known polyphenisms have this kind of indirect envi- allometric relationships, and it is unlikely that this can be ronmental sensitivity, and this is also the case for many phe- achieved without a qualitative switch-like change in the de- notypes that exhibit continuous phenotypic plasticity, or reac- velopmental processes that give rise to the phenotypes tion norms. As noted above, many polyphenisms are nothing (Nijhout and Wheeler 1982 1996). more than reaction norms that are sparsely sampled. It would be interesting to find out whether there are any cases of adap- tive phenotypic plasticity in animals in which development A CASE STUDY: ONTHOPHAGUS TAURUS of the relevant trait is directly sensitive to the environmental variable or whether all cases are mediated by evolved inte- Here I present a case study of the development and evolution grated systemic processes, as in the case of polyphenisms. of a polyphenism that illustrates many of the features out- The discrete alternative morphs of a polyphenisms arise if lined above. Males of the dung beetle Onthophagus taurus there is an unambiguous hit or miss of hormone secretion have a horn length polyphenism. Small males are essentially and hormone-sensitive period in different individuals. When hornless, whereas large males have well-developed cephalic there is partial overlap between a hormone pulse and a sen- horns (Fig. 6). Horn length varies allometrically with body sitive period, it is possible to get an intermediate phenotype. size, but the allometry is highly nonlinear. This results in a Such intermediates can be artificially produced by manipu- bimodal distribution of horn sizes, even though body size is lating the timing of a hormone pulse, as illustrated in Figure normally distributed (Fig. 7). Each of the two horn morphs 4. In this way it is possible to generate a range of intermedi- is an adaptation to differences in body size. Males defend ate phenotypes from what in nature is an invariant diphen- tunnels dug by females and use their horns to combat other ism. In almost all polyphenisms it is possible to obtain a males for access to females (Emlen 1997a). Males with large smoothly continuous range of intermediates either by envi- body and horn sizes inevitably win contests when matched ronmental or physiological manipulation. This continuity against males with smaller body and horn sizes (Moczek and suggests that the developmental mechanisms that give rise to Emlen 2000). One might believe that this should result in an

Fig. 6. Head width allometry in Pheidole bicarinata. Allometries of soldiers and workers are not continuous, so that it is possible to have two distinct morphologies in animals of the same body size (Wheeler and Nijhout 1983).

16 EVOLUTION & DEVELOPMENT Vol. 5, No. 1, January–February 2003

Fig. 7. Horn polyphenism in males of the beetle Onthophagus taurus. Horns are sigmoidally allo- metric with body size. As a result, horn size has a bimodal frequency distribution (inset). Beetles that metamorphose at a large body size develop large cephalic horns, whereas those that metamor- phose at small body sizes are virtually hornless. evolutionary escalation of body size, but this is not the case at some distance below the defending male and thus gain ac- because body size in this beetle is not heritable (Emlen 1994; cess to the female. The alternative mating tactics of large- and Moczek 1998; Moczek and Emlen 1999). Body size is deter- small-bodied males result in a divergent selection on horn mined entirely by the size and quality of the food supply with size. Large males benefit from having the largest possible which a mother provisions a given egg. When a runs horns, because these help them win combats. Small males out of food, it almost immediately begins metamorphosis to benefit from having the smallest possible horns, because horns the adult (Shafiei et al. 2001). Variation in food supply and get in the way of their sneaking tactic. This divergent selec- in the nutritive quality of that food are the environmental tion on relative horn size has resulted in a sharply sigmoidal variables that determine the range and distribution of body body size–horn size allometry. Horn size in this beetle is a sizes. good example of adaptive phenotypic plasticity with two Males that happen to be small are at a competitive disad- adapted extremes. The allometry of horns is effectively a re- vantage with large males, and this has led to the evolution of action norm of horn size on nutrition. an alternative mating tactic in small males (Emlen 1997a; The threshold body size at which the transition from horn- Moczek and Emlen 2000). These small males do not attempt less to horned males occurs is itself a plastic trait. When lar- to fight but either attempt to sneak past a defending male or vae are fed on food of low nutritive quality, mean body size actually dig their own tunnels that intersect defended tunnels is smaller and the transition to horn development occurs at a

Nijhout Development and evolution of adaptive polyphenisms 17 smaller body size than it does in larvae fed on a nutrient-rich tion. This would represent a physiological co-option of a pre- diet (Emlen 1997b). existing regulatory mechanism. The subsequent suppression The developmental regulation of horn expression has sev- of horns in small males required the evolution of a size-sensing eral components. Larvae of Onthophagus must somehow as- mechanism in the final larval instar. Some insects monitor sess the presumptive body size of the adult, so that horn devel- their body size by means of abdominal stretch reception opment can be suppressed or initiated, as appropriate. This (Nijhout 1994), but it is not known whether this is the case assessment occurs in about the middle of the last larval instar here. Our current hypothesis is that this size-sensing mecha- (Emlen and Nijhout 1999) and coincides with a brief peak of nism during the larval stage regulates the program of juve- ecdysone secretion that occurs in females and presumptive nile hormone level during the prepupal stage. Juvenile hor- hornless males, but not in presumptive horned males. Treatment mone normally declines just before the prepupal stage and is of presumptive hornless animals with during absent at the time that ecdysone-stimulated cell division oc- this critical period induces subsequent horn development. curs in the horns. It is possible that in small males the decline Horns develop during the prepupal stage from small of juvenile hormone is delayed, so that it is still above thresh- imaginal disk-like clusters of cells on the head. In horned old and is able to inhibit ecdysone-stimulated horn growth. males these cells proliferate extensively during a very brief period of time, forming an elongated projection that will de- REFERENCES velop into the adult horn during metamorphosis. Similarly focused cell divisions do not occur in presumptive hornless Cook, C. D. K. 1968. Phenotypic plasticity with particular reference to three amphibious plant species. In V. Heywood (ed.). Modern Methods males. There is a second period of hormone sensitivity in the in Plant Taxonomy. Academic Press, London, pp. 97–111. prepupal stage, immediately before the time of horn differ- Dingle, H., and Winchell, R. 1997. Juvenile hormone as a mediator of plas- entiation. If prepupae are treated with juvenile hormone dur- ticity in insect life histories. Arch. Insect Biochem. Physiol. 35: 359–373. Emlen, D. J. 1994. Environmental control of horn length dimorphism in ing this time, cell proliferation in the presumptive horns is the beetle Onthopagus acuminatus (Coleoptera: Scarabaeidae). Proc. suppressed (Emlen and Nijhout 2001). The presence of juve- Roy. Soc. London Ser. B. 256: 131–136. nile hormone during this second sensitive period thus inhib- Emlen, D. J. 1997a. Alternative mating tactics and male dimorphism in the horned beetle Onthophagus acuminatus (Coleoptera: Scarabaeidae). its horn development, which implies that normal horn devel- Behav. Ecol. Sociobiol. 41: 335–341. opment requires that juvenile hormone is absent at this time. Emlen, D. J. 1997b. Diet alters male horn allometry in the beetle Ontho- Although the pathway of horn regulation has not yet been phagus acuminatus (Coleoptera: Scarabaeidae). Proc. R. Soc. Lond. Ser. B 264: 567–574. fully elucidated, it appears at present that it depends on two Emlen, D. J., and Nijhout, H. F. 1999. Hormonal control of male horn hormone-sensitive periods. A brief period of ecdysone secre- length dimorphism in the dung beetle Onthophagus taurus (Coleoptera: tion during the first hormone-sensitive period is regulated by Scarabaeidae). J. Insect Physiol. 45: 45–53. Emlen, D. J., and Nijhout, H. F. 2000. The development and evolution of a size assessment mechanism and may interact with juvenile exaggerated morphologies in insects. Annu. Rev. Entomol. 45: 661–708. hormone to program horn development in the prepupal Emlen, D. J., and Nijhout, H. F. 2001. Hormonal control of male horn stage. Horn development in the prepupa is, in turn, regulated length dimorphism in the dung beetle Onthophagus taurus (Coleoptera: Scarabaeidae): a second critical period of sensitivity to juvenile hor- during a second hormone-sensitive period during which ju- mone. J. Insect Physiol. 47: 1045–1055. venile hormone controls whether or not epidermal cells in Gomulkiewicz, R., and Kirkpatrick, M. 1992. and the the horn-forming region of the head will undergo cell divi- evolution of reaction norms. Evolution 46: 390–411. Harvell, C. D. 1990. The ecology and evolution of inducible defenses. Q. sions. Cell division in the insect epidermis is stimulated by Rev. Biol. 65: 323–340. ecdysone, so the simplest way in which juvenile hormone Lively, C. M. 1986. Canalization versus developmental conversion in a spa- can exert its inhibiting effect is by suppressing the expres- tially variable environment. Am. Nat. 128: 561–572. Moczek, A. P. 1998. Horn polyphenism in the beetle Onthophagus taurus: sion of ecdysone receptors in the appropriate epidermal larval diet quality and plasticity in parental investment determine adult cells. body size and male horn morphology. Behav. Ecol. 9: 636–641. The evolutionary origin of horn polypenism in Onth- Moczek, A. P., and Emlen, D. J. 1999. Proximate determination of male horn dimorphism in the beetle Onthophagus taurus (Coleoptera: Scar- ophagus thus involved two events: the origin of cephalic horns abaeidae). J. Evol. Biol. 11: 27–37. in males and the suppression of horn development in males Moczek, A. P., and Emlen, D. J. 2000. Male horn dimorphism in the scarab that are smaller than a threshold size. Both events are regu- beetle Onthophagus taurus: do alternative reproductive tactics favour alternative phenotypes? Anim. Behav. 59: 459–466. lated by hormones. Epidermal growth requires ecdysone, and Moran, N. A. 1992. The evolutionary maintenance of alternative pheno- the localized epidermal proliferation that results in horn de- types. Am. Nat. 139: 971–989. velopment requires the localized expression of ecdysone Nijhout, H. F. 1994. Insect Hormones. Princeton University Press, Prince- ton, NJ. receptors. In addition, a period of ecdysone secretion is re- Nijhout, H. F. 1999. Control mechanisms of polyphenic development in in- quired to induce cell proliferation. Insofar as ecdysone is sects. Bioscience 49: 181–192. used to regulate cell division and in many Nijhout, H. F., and Davidowitz, G. 2002. Developmental perspectives on phenotypic instability, canalization and fluctuating asymmetry. In: Po- other tissues during metamorphosis, it is possible that horn lak, M. (ed.). Developmental Instability: Causes and Consequences. Ox- development captured a preexisting peak of ecdysone secre- ford University Press, New York (in press).

18 EVOLUTION & DEVELOPMENT Vol. 5, No. 1, January–February 2003

Nijhout, H. F., and Wheeler, D. E. 1982. Juvenile hormone and the physi- Via, S., Gomulkiewicz, R., De Jong, G., Scheiner, S. M., Schlichting, C. D., ological basis of insect polyphenisms. Q. Rev. Biol. 57: 109–133. and Van Tienderen, P. H. 1995. Adaptive phenotypic plasticity: consen- Nijhout, H. F., and Wheeler, D. E. 1996. Growth models of complex allom- sus and controversy. Trends Ecol. Evol. 10: 212–217. etries in insects. Am. Nat. 148: 40–56. Via, S., and Lande, R. 1985. Genotype-environment interaction in the evo- Schlichting, C. D., and Pigliucci, M. 1998. Phenotypic Evolution. A Reac- lution of phenotypic plasticity. Evolution 39: 505–522. tion Norm Approach. Sinauer Associates, Sunderland, MA. West-Eberhard, M. J. 1989. Phenotypic plasticity and the origins of diver- Shafiei, M., Moczek, A. P., and Nijhout, H. F. 2001. Food availability con- sity. Annu. Rev. Ecol. System. 20: 249–278. trols the onset of metamorphosis in the dung beetle Onthophagus taurus Wheeler, D. E. 1986. Developmental and physiological determinants of (Coleoptera: Scarabaeidae). Physiol. Entomol. 26: 173–180. caste in social Hymenoptera: evolutionary implications. Am. Nat. 128: Talbot, W. S., Swyryd, E. A., and Hogness, D. S. 1993. Drosophila tissues 13–34. with different metamorphic responses to ecdysone express different Wheeler, D. E., and Nijhout, H. F. 1983. Soldier determination in Pheidole ecdysone receptor isoforms. Cell 73: 1323–1337. bicarinata: effect of methoprene on caste and size within caste. J. Insect Truman, J. W., Talbot, W. S., Fahrbach, S. E., and Hogness, D. S. 1994. Physiol. 29: 847–854. Ecdysone receptor expression in the CNS correlates with stage-specific Windig, J. J. 1994. Reaction norms and the genetic basis of phenotypic plas- responses during Drosophila and Manduca development. Development ticity in the wing pattern of the butterfly Bicyclus anynana. J. Evol. Biol. 120: 219–234. 7: 665–695.